Television – Special applications – Observation of or from a specific location
Reexamination Certificate
1999-07-28
2002-09-03
Britton, Howard (Department: 2613)
Television
Special applications
Observation of or from a specific location
C348S169000
Reexamination Certificate
active
06445409
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a monitoring technique using an image input device such as a camera, or more in particular to a moving object recognition method for automatically detecting a moving object that has entered the imaging view field in distinction from other various objects not to be monitored in the video signal obtained, and an apparatus for automatically tracking and monitoring a moving object detected using the method.
An object tracking and monitoring apparatus using an image input device such as a camera has been widely used. In recent years, however, a monitoring system has come to be required in which instead of a manned monitoring for detecting and tracking a moving object such as a human being, an automobile or a ship on the sea that has entered the view field of monitoring while watching the image displayed on the monitor, an intruding object can be automatically detected and tracked by an image signal input from an image input device such as a camera and a predetermined information or an alarm is given.
In a method for realizing such a system, the input image obtained from an image input device is compared with a reference background image (i.e. the image not including the object to be detected), and the difference in intensity is determined for each pixel, and an area with a large intensity difference is detected as an object. This method is called a subtraction method and has been widely used.
SUMMARY OF THE INVENTION
An application of the object tracking method using the subtraction method described above is the invention disclosed in U.S. Ser. No. 09/078,521 (EP 98108707.5) assigned to the same assignor as the assignee of the present application. The tracking method using the subtraction method poses the problems of a splitting in which a single object is observed and tracked temporarily as a plurality of objects and a coupling in which a plurality of objects are observed and tracked as a single object temporarily. These phenomena occur when detecting an object having a intensity near to that of the background image. In the invention disclosed in U.S. Ser. No. 09/078,521 (EP 98108707.5) described above, the change of connection state of a detected object between successive frames is classified into five states including appearance, single coupling, coupling, disappearance and separation, and these connection state changes are stored sequentially together with information such as the center of gravity or the detection range or the size of the object. Further, the inconveniences of the splitting or coupling are corrected based on the connection state change and information of the detected object in each frame thereby to realize an accurate tracking of an object.
As described above, the subtraction method is for detecting an object based on the difference for each pixel between the input image from an image input device such as a camera and the background image set in advance. In other words, the subtraction method is not selective in objects of selection. Another problem of the subtraction method, therefore, is that moving objects other than the object to be detected, such as trees and ocean waves in the view field (objects not required for detection) are detected undesirably.
More specifically, trembling of plants, fluttering of paper, cloth or the like and the intrusion of small living creatures or fallen leaves in the wind on the land and the motion of waves crushing against the wharf, floating trees on the water surface, the sun entering or leaving the clouds, the instantaneous intensity change of the lighthouse or the vehicle light, are not required to be detected in many cases. The conventional subtraction method poses the problem that an accurate tracking of a moving object is impossible in a scene having such an unrequired moving object.
The object of the present invention is to obviate these problems and to provide a reliable method of recognizing a moving object and a reliable apparatus for tracking and monitoring a moving object in which an object to be detected and other objects are distinguished from each other.
In order to achieve this object, according to one aspect of the invention, there is provided a moving object recognition method for recognizing a specified moving object in sequentially input image signals in distinction from other objects, comprising the steps of sequentially detecting at least one moving object in the sequentially input image signals by a subtraction method, sequentially storing, for each input image signal, in a memory at least one characteristic physical quantity which the moving object has, detecting at least one locus of the moving object based on characteristic physical quantities in the input image signals, storing in a memory characteristic physical quantities of the detected locus in such a manner that the characteristic physical quantities are associated with each other as a series of characteristic physical quantities of the moving object, previously setting a test or criterion of determining whether the moving object belongs to the specified moving object or belongs to objects other than the specified moving object, and determining or classifying from the series of characteristic physical quantities, whether or not said moving object is the specified moving object, based on the determining criterion.
According to an embodiment, the criterion or test described above includes the steps of calculating the time during which the locus of a moving object continues to exist from a series of time and positions at which the moving object is detected as included in the correlated series of characteristic physical quantities (referred to as physical quantities) stored in memory, and determining a moving object with the calculated time longer than a predetermined time as an object to be recognized while determining a moving object with the calculated time not longer than the predetermined time as an object not to be recognized.
Preferably, the locus of the detected moving object is determined using more than two input image signal frames, and the time during which the locus continues to exist is calculated using the number of continuous image signal frames at the time when the locus is detected.
According to another embodiment, the criterion or test described above includes the steps of calculating the variations of the change of the position of the moving object from a series of time and positions at which the moving object is detected as included in the correlated series of the physical quantities stored in memory, and determining a moving object with the calculated position change variation smaller than a predetermined value as an object to be recognized and a moving object with the calculated position change variation not smaller than the predetermined value as an object not to be recognized.
According to still another embodiment, the criterion or test described above includes the steps of calculating the maximum size of a detected moving object from the area thereof included in the correlated series of physical quantities stored in memory, and determining a moving object with the calculated maximum size larger than a predetermined value as an object to be recognized and a moving object with the maximum size not larger than the predetermined value as an object not to be recognized.
According to an embodiment, the area included in the series of physical quantities of each moving object is calculated from a circumscribed polygon about the area occupied by the moving object.
According to another embodiment, the area included in the information series of each moving object is calculated from the area occupied by the moving object.
According to another embodiment, the criterion or test includes the steps of calculating the minimum area of each of the moving objects from the correlated series of physical quantities stored in memory and classifying or determining a moving object with the calculated minimum area smaller than a predetermined area as an object to be recognized and a moving
Ito Wataru
Ueda Hirotada
Yamada Hiromasa
Antonelli Terry Stout & Kraus LLP
Britton Howard
Hitachi Denshi Kabushiki Kaisha
LandOfFree
Method of distinguishing a moving object and apparatus of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of distinguishing a moving object and apparatus of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of distinguishing a moving object and apparatus of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2907351